Who Holds the Purse Strings?

Abstract

Couples in Australian households make financial decisions in two distinct modes — jointly as a couple unit, or individually with one partner taking the lead — and the choice between modes correlates with both partner satisfaction and subsequent separation hazard in ways that family-of-origin and household-composition controls do not fully explain. The behavioural-finance literature has documented these correlations across multiple panel datasets; what has not been clearly identified is whether the decision-mode choice causes relational decline, reflects it, or is jointly determined by an unobserved third factor.

The conceptual framework treats intra-family financial decision-making as a behavioural and relational mechanism rather than a static household characteristic. The paper separates day-to-day financial control, investment decision-making, income contribution, satisfaction, and separation risk as distinct constructs — rather than treating household financial decision-making as one broad variable. The behavioural question is which combinations of decision power are protective, neutral, or destabilising for couples.

The empirical setting is the HILDA panel, restricted to coupled households with non-missing items in the financial-decision-making module and the partner-satisfaction battery. The analytic samples are 93,826 observations for the day-to-day-decisions model and 124,907 observations for the joint-decisions model. The existing analysis applies the classical identification stack: OLS linear-probability models with household and regional fixed effects, logit and probit estimation for the binary decision-mode outcomes, random-effects panel regression for the within-household dynamics, Oaxaca-Blinder decomposition of the gender-asymmetric decision-mode pattern, and Cox proportional-hazards modelling for separation duration. The existing classical evidence already shows that females are less likely to make investment decisions themselves and more likely to make day-to-day decisions themselves; both genders making decisions themselves report lower partner satisfaction.

We propose extending the existing analysis with a Causal Forest CATE estimator that would recover the heterogeneous effect of decision-making mode on partner satisfaction across the gender × income-contribution-share strata. The proposed extension, once implemented, would identify whether the satisfaction penalty is concentrated in the 50–75 percent income-contribution band and supply a couples-counselling-segment screening rule.

Data & Methods